Data Sources and Technical Notes for the 2008 Data Update

 Data Sources:

  • Data from NORC come the author's analyses of the 1998-2006 General Social Survey.
  • Data from ISR come from the author's analyses of the 1998-2004 National Election Studies.
  • Data from Gallup were based on published data from the Gallup organization.
  • Data from CBS/NYT come from the author's analyses of CBS/NYT data available through ICPSR.

Technical Notes:

  • All data reported are for respondents who are 21 years and older, except for the data from Gallup, which are for respondents 18 years and older.
  • Sample sizes for white respondents were all in excess of 1,000 cases, unless specified in the notes to the table (see the Question Wording and Notes section in each table for details.)
  • Sample sizes for African American respondents were substantially smaller. In cases where the sample size fell below 100 cases, the results are shown in parentheses (). For most of the NES data, the sample sizes for African American respondents were just over 100 cases. The sample sizes for the NORC and Gallup data ranged from several hundred to nearly 1,000 (in the case of Gallup).
  • There have been shifts in mode of administration over time. Recent Gallup data come from telephone surveys; CBS/NYT data are based on telephone surveys; and NORC's General Social Survey continues to be face-to-face. The NES has used some of both (see next two points for details about the 1998 and 2000 NES). Unless specified in the "Question Wording and Notes" section, data can be presumed to be from face-to-face interviews.
  • Most of the respondents in the 1998 NES were interviewed by telephone (77 percent), with a small sample by face-to-face (33 percent). Area probability sampling was used to select the sample and interviewers went to the households and obtained telephone numbers, which were then used to conduct the telephone interviews. Because mode of administration was not randomly assigned, the data we present in these tables were from the full sample of 1998 NES respondents.
  • The National Election Study in 2000 included a split-ballot mode experiment. For reasons of comparability to previous time points, and because the mode of administration was randomly assigned to respondents (unlike the 1998 NES), our analysis includes only respondents who were interviewed in the face-to-face mode.
  • NORC data for 2004 and 2006 are weighted data as necessitated by changes in sampling design that year; all other data are unweighted.
  • Due to rounding error, results for some questions do not sum to 100%.